Some fundamentals of numerical weather and climate prediction Robert Fovell Atmospheric and Oceanic Sciences University of California, Los Angeles

Slides:



Advertisements
Similar presentations
Chapter 13 – Weather Analysis and Forecasting
Advertisements

ESC Global Climate Change Chapter 5
Section 2: The Planetary Boundary Layer
Computational Fluid Dynamics - Fall 2013 The syllabus CFD references (Text books and papers) Course Tools Course Web Site:
The Problem of Parameterization in Numerical Models METEO 6030 Xuanli Li University of Utah Department of Meteorology Spring 2005.
Hurricanes and climate ATOC 4720 class22. Hurricanes Hurricanes intense rotational storm that develop in regions of very warm SST (typhoons in western.
Cold Fronts and their relationship to density currents: A case study and idealised modelling experiments Victoria Sinclair University of HelsinkI David.
1 Use of Mesoscale and Ensemble Modeling for Predicting Heavy Rainfall Events Dave Ondrejik Warning Coordination Meteorologist
2012: Hurricane Sandy 125 dead, 60+ billion dollars damage.
DARGAN M. W. FRIERSON DEPARTMENT OF ATMOSPHERIC SCIENCES DAY 16: 05/20/2010 ATM S 111, Global Warming: Understanding the Forecast.
Climate modeling Current state of climate knowledge – What does the historical data (temperature, CO 2, etc) tell us – What are trends in the current observational.
Chapter 9: Weather Forecasting Acquisition of weather information Acquisition of weather information Weather forecasting tools Weather forecasting tools.
It has been said that “weather is an initial value problem, whereas climate is a boundary-value problem.” What is meant by this statement? Is this statement.
Natural Environments: The Atmosphere
Earth Systems Science Chapter 6 I. Modeling the Atmosphere-Ocean System 1.Statistical vs physical models; analytical vs numerical models; equilibrium vs.
Climate models – prediction and projection Nils Gunnar Kvamstø Geophysical Department University of Bergen.
AOS 100: Weather and Climate Instructor: Nick Bassill Class TA: Courtney Obergfell.
AOS 100: Weather and Climate Instructor: Nick Bassill Class TA: Courtney Obergfell.
A History of Modern Weather Forecasting. The Stone Age Prior to approximately 1955, forecasting was basically a subjective art, and not very skillful.
A Voyage of Discovery Physical oceanography Instructor: Dr. Cheng-Chien LiuCheng-Chien Liu Department of Earth Sciences National Cheng Kung University.
Introduction to Numerical Weather Prediction and Ensemble Weather Forecasting Tom Hamill NOAA-CIRES Climate Diagnostics Center Boulder, Colorado USA.
Weather Forecasting - II. Review The forecasting of weather by high-speed computers is known as numerical weather prediction. Mathematical models that.
UNDERSTANDING TYPHOONS
SC.912.E.7.5 Predict future weather conditions based on present observations and conceptual models and recognize limitations and uncertainties of such.
Things to look for on the weather maps Visible and IR satellite images (& radar too): Look at cloud movements and locations - do they correlate with what.
Lapse Rates and Stability of the Atmosphere
Evaporation Slides prepared by Daene C. McKinney and Venkatesh Merwade
NUMERICAL WEATHER PREDICTION K. Lagouvardos-V. Kotroni Institute of Environmental Research National Observatory of Athens NUMERICAL WEATHER PREDICTION.
Lecture Oct 18. Today’s lecture Quiz returned on Monday –See Lis if you didn’t get yours –Quiz average 7.5 STD 2 Review from Monday –Calculate speed of.
Atmospheric pressure and winds
The simplest theoretical basis for understanding the location of significant vertical motions in an Eulerian framework is QUASI-GEOSTROPHIC THEORY QG Theory:
Evaporation What is evaporation? How is evaporation measured? How is evaporation estimated? Reading: Applied Hydrology Sections 3.5 and 3.6 With assistance.
Sources of Error in NWP Forecasts or All the Excuses You’ll Ever Need Fred Carr COMAP Symposium 00-1 Monday, 13 December 1999.
Computational Fluid Dynamics - Fall 2003 The syllabus Term project CFD references (Text books and papers) Course Tools Course Web Site:
Simulating Supercell Thunderstorms in a Horizontally-Heterogeneous Convective Boundary Layer Christopher Nowotarski, Paul Markowski, Yvette Richardson.
Some fundamentals of numerical weather prediction (for ATM 562) Robert Fovell Atmospheric and Environmental Sciences University at Albany, SUNY
Ch 9.8: Chaos and Strange Attractors: The Lorenz Equations
ADVENTURE IN SYNOPTIC DYNAMICS HISTORY
Chapter 9: Weather Forecasting Acquisition of weather information Acquisition of weather information Weather forecasting tools Weather forecasting tools.
Future Climate Projections. Lewis Richardson ( ) In the 1920s, he proposed solving the weather prediction equations using numerical methods. Worked.
Modeling the Atmospheric Boundary Layer (2). Review of last lecture Reynolds averaging: Separation of mean and turbulent components u = U + u’, = 0 Intensity.
Sensitivity Analysis of Mesoscale Forecasts from Large Ensembles of Randomly and Non-Randomly Perturbed Model Runs William Martin November 10, 2005.
What set the atmosphere in motion?
Convective Roll Effects on Sea Breeze Fronts
ATS/ESS 452: Synoptic Meteorology
Weather forecasting by computer Michael Revell NIWA
Synoptic Scale Balance Equations Using scale analysis (to identify the dominant ‘forces at work’) and manipulating the equations of motion we can arrive.
AOS 100: Weather and Climate Instructor: Nick Bassill Class TA: Courtney Obergfell.
AOSS 401, Fall 2007 Lecture 21 October 31, 2007 Richard B. Rood (Room 2525, SRB) Derek Posselt (Room 2517D, SRB)
Observed Structure of the Atmospheric Boundary Layer
AOSS 401, Fall 2006 Lecture 18 October 24, 2007 Richard B. Rood (Room 2525, SRB) Derek Posselt (Room 2517D, SRB)
Page 1© Crown copyright Modelling the stable boundary layer and the role of land surface heterogeneity Anne McCabe, Bob Beare, Andy Brown EMS 2005.
ATS/ESS 452: Synoptic Meteorology Friday 08 January 2016 Review Material Overview of Maps Equations of Motion Advection Continuity.
AOSS 401, Fall 2006 Lecture 16 October 19, 2007 Richard B. Rood (Room 2525, SRB) Derek Posselt (Room 2517D, SRB)
Computational Fluid Dynamics - Fall 2007 The syllabus CFD references (Text books and papers) Course Tools Course Web Site:
Figures from “The ECMWF Ensemble Prediction System”
ThermodynamicsM. D. Eastin We need to understand the environment around a moist air parcel in order to determine whether it will rise or sink through the.
Tropical dynamics and Tropical cyclones
Numerical Weather Forecast Model (governing equations)
Technology on the Cutting Edge of Weather Research and Forecasting
Grid Point Models Surface Data.
Robert Fovell Meteorology – Lecture 23 Robert Fovell
How do models work? METR 2021: Spring 2009 Lab 10.
Dynamical Models - Purposes and Limits
The Stone Age Prior to approximately 1960, forecasting was basically a subjective art, and not very skillful. Observations were sparse, with only a few.
Modeling the Atmos.-Ocean System
How will the earth’s temperature change?
Why NWS Forecasts go Awry and Steps to Improve Them
Computational Fluid Dynamics - Fall 2001
What is Weather?.
Presentation transcript:

Some fundamentals of numerical weather and climate prediction Robert Fovell Atmospheric and Oceanic Sciences University of California, Los Angeles

Equations Navier-Stokes equations –Newton’s 2 nd law: real & apparent forces –Turbulence and mixing 1 st law of thermodynamics Ideal gas law Continuity equation Clausius-Clapeyron equation Discretize in space and time

Domain Local, regional or global in scale

Domain Local, regional or global in scale Discretize into grid volumes –virtual internal walls –boundary conditions Initialize each volume Make some forecasts…

Extrapolation Equations predict tendencies –Previous forecasts used to recalculate tendencies –Initial forecasts start with observations but subsequent forecasts based on forecasts –Success depends on quality of initialization and accuracy of tendencies

A simplistic example: temperature in your backyard

The model forecast will consider an enormous number of factors to estimate present tendency to project future value

In the absence of such information, you simply guess… and wait to see how good your guess was.

The model does not wait. It uses each forecast to recalculate the tendencies to make the next forecast

Verification of your forecast: OK since you didn’t project out too far

Forecast time steps cannot be too long. Tendencies have the tendency to CHANGE.

Forecasts will reflect… Radiative processes Surface processes Cloud development and microphysics Advection and mixing Convergence and divergence Ascent and subsidence Sea-breezes, cold and warm fronts Cyclones, anticyclones, troughs, ridges

Model resolution Things to try to resolve –Clouds, mountains, lakes & rivers, hurricane eyes and rainbands, fronts & drylines, tornadoes, much more What isn’t resolved is subgrid At least 2 grid boxes across a feature for model to even “see” it… and at least 6 to render it properly

Wave-like features are ubiquitous.

Models sample wave only at grid points. Example: 4 points across each wave.

Models “connect the dots”.

Suppose we have only 3 points across the wave… or just 2 points…

These do not look much like the actual wave at all. With only 2 points, wave may be invisible.

High resolution Hurricane Katrina satellite picture = composed of 1 km pixels. High resolution model = possibly accurate, definitely expensive, and extremely time-consuming

As model resolution increases, we see more, but have to DO more. Plus, the time step has to decrease, to maintain linear and nonlinear stability

Compromising on the resolution

30 km resolution

At what point would we not be able to tell that’s a hurricane if we had not known it from the start?

This is the world as seen by global weather models not so long ago, and many climate models today.

Outlook is good… Models are improving –Faster computers –More and better input data (satellites) –Better ways of using those data –Progressively better numerical techniques –High resolution models… so less is missed (subgrid) –… but some things we can never capture

Parameterization Parameterization = an attempt to represent what we cannot see, based largely on what we can Parameterizations in a typical weather/climate model include - and are NOT limited to… –Boundary layer processes and subgrid mixing –Cloud microphysics [resolve cloud, can’t resolve drops] –Convective parameterizations [can’t resolve clouds] –Surface processes [heat, moisture fluxes] –Subsurface processes [soil model, ocean layers] –Radiative transfer, including how radiation interacts with clouds Here is an example…

Another view of Katrina... But focusing on roll clouds. Roll clouds accomplish boundary layer mixing.

How roll clouds form Consider the sun warming the land during the day

How roll clouds form With uneven heating, wind and vertical wind shear, roll-like circulations can start You can simulate roll formation in a higher resolution model, like DTDM

How roll clouds form As the land warms up, the rolls get deeper The rolls mix heat vertically –air is a lousy conductor Also mixing momentum, moisture

How roll clouds form If conditions are favorable, clouds will form above the roll updrafts These make the rolls visible

We see these roll clouds on satellite pictures In a sense, being able to simulate the roll clouds means we’ve done many things well - - radiation, winds, mixing, saturation processes

Parameterizing mixing… In many models, however, the clouds and the mixing that created them are subgrid But that mixing is important. It influences structure and stability of the atmosphere. If we cannot resolve it, we must parameterize it Weather/climate models are full of parameterizations, each a potential model shortcoming, each a possible source of problems, of error, of uncertainty regarding the future

Birth of Numerical Weather and Climate Prediction

Prof. Cleveland Abbe “There is a physical basis for all meteorological phenomena. There are laws of mechanics and heat that apply to the atmosphere, and as fast as we acquire the ability to discover and reason out their consequences, we shall perceive that LAW and ORDER prevail in all the complex phenomena of the weather and the climate.” (1901) First head of US Weather Bureau

Prof. Vilhelm Bjerknes 1904 vision on weather prediction Lamented the “unscientific” basis of meteorology Goal: to make meteorology a more exact science … by making predictions

Bjerknes’ vision Two key ingredients: –Sufficiently accurate knowledge of the state of the atmosphere at the initial time –Sufficiently accurate knowledge of the physical laws that govern how the atmospheric state evolves Identified 7 fundamental variables -- T, p, density, humidity, and three wind components -- and the equations that calculated their tendencies These are nasty equations, without simple solutions. Only numerical methods could be brought to bear on them.

Max Margules Austrian meteorologist Tried to predict surface pressure using continuity equation… and found it could produce very poor results In 1904, he declared this would be impossible and that weather forecasting was “immoral and damaging to the character of a meteorologist.”

Lewis Fry Richardson Among first to try to solve the weather/climate equations numerically Invented many concepts still used today Made the first numerical weather forecast Prediction was horribly wrong

Richardson’s technique Richardson laid out a grid, collected his observations, created novel numerical approximations for his equations and crunched his numbers, by hand and slide rule Actually, it was a hindcast that took laborious computations using old, tabulated data

Richardson’s grid Richardson used 5 vertical levels, including surface

Richardson’s forecast He predicted a surface pressure rise of 145 mb (about 14.5%) in only 6 h In reality the pressure hardly changed at all First forecast = first forecast failure What went wrong?

Richardson EXTRAPOLATED too far. His technique made monsters out of meaningless oscillations that happen as air wiggles up and down in a stable atmosphere.

Richardson’s book Richardson revealed the details of his blown forecast in “Weather Prediction by Numerical Process”, published in Undaunted, he imagined his pencil and paper technique applied to the entire global atmosphere… –In a huge circular ampitheatre in which human calculators would do arithmetic and, guided by a conductor at the center, pass results around to neighbors

Richardson’s forecasting theater

After Richardson Richardson’s confidence was not unfounded. Mathematicians and meteorologists realized the flaws of his techniques The digital computer was created The meteorological observation network was expanding rapidly It seemed only a matter of time until the vision of Abbe, the goal of Bjerknes, the dream of Richardson, became a reality… Then in 1962, Ed Lorenz did his little experiment…

Prof. Ed Lorenz MIT professor Landmark 1962 paper “Deterministic Nonperiodic Flow” Led to “chaos theory” and dynamical systems Coined “butterfly effect”

The Lorenz Experiment Lorenz’ model wasn’t a weather model, and didn’t even have grid points 3 simple equations, which can describe fluid flow in a cylinder with heated bottom and cooled top He called his variables X, Y and Z

The Lorenz Experiment X indicated the magnitude and direction of the overturning motion As X changed sign, the fluid circulation reversed

The Lorenz Experiment Y was proportional to the horizontal T gradient

The Lorenz Experiment And Z revealed the fluid’s stability

The Lorenz Model Three simple equations But in important ways they were like the equations we use in weather forecasting –They are coupled –They are nonlinear

Simulation similar to Lorenz’ original experiment. X = circulation strength & magnitude.

The model was started with unbalanced initial values for X, Y and Z, creating a shock. The model was seeking a suitable balance.

Next, a spin-up period with swings that grow until…

The fluid chaotically shifts from CW to CCW circulations in an nonperiodic fashion

This was Lorenz’ discovery… sensitive dependence on initial conditions

Dependence on initial conditions Caused by nonlinear terms Even if model is perfect, any error in initial conditions means forecast skill decreases with time Reality: models are far from perfect Long range weather prediction is impossible Lorenz: “We certainly had been successful at doing that anyway and now we had an excuse.”

Lorenz attractor 3D space with coordinates being the Lorenz variables, X Y and Z Predicted values can be plotted as a point in this space Start with an initial conditions that are slightly different…, … they diverge in this space Birth of chaos theory Poorly named: chaos ≠ random

If we cannot produce accurate weather forecasts for next week, how can we trust climate forecasts for next decade, or next century? The simple answer is: weather is not climate

Note weather is different but climate is similar.

Summary Weather and climate models = great tools for understanding our present and predicting our future Models initialized with data to compute tendencies, via extrapolation Models have limitations –Resolution –Unresolvable features and processes –Incomplete or erroneous initial conditions New forecasts based on previous ones, so error grows Limit to weather predictability (Lorenz) Weather ≠ climate

[end]